When customers need to expand an Azure Greenplum cluster—such as doubling disk space from 2TB to 4TB—there are three primary approaches to achieve this.
GPGrow Utility: Used to increase the disk size for segment instances, triggering an automatic restart for segments to complete the process.
Downtime: The cluster needs to be stopped briefly. Downtime is mainly determined by detaching and reattaching instances.
Marketplace Limitation: Part of the cluster may be managed by Azure Marketplace templates, making in-place expansion inconsistent.
Workarounds:
Create a new cluster of the desired size, then copy data from the old cluster.
Use GPCopy to move data between clusters, or use GPBackup/GPRestore for backup/restore operations.
Network speed directly impacts total move time.
Notes:
GPGrow does not support increasing master instance disk size.
Azure only permits increasing disk sizes; this operation cannot be conducted online.
The utility handles 3 data disks per segment instance, with size specified in GB.
Add Compute/RAM: Expanding CPU or memory resources per segment instance.
Concurrency: Ideal for scenarios requiring higher concurrency (more users or parallel operations).
This does not directly involve changing disk capacity.
Redistributing Data: Adding more segments requires redistributing existing data across the cluster.
GPExpand Utility: Mandatory after segment expansion.
Complexity & Limitations:
Least common method due to complexity.
Reduces the data-to-resource ratio but can lower parallelism.
Virtualized marketplace clusters lack flexibility for this approach; new VMs and custom networking/routing are required.
Can be a costly operation considering resource requirements.
Product Version: 6.21
If needed, please contact GPDB support for more detailed information on resizing.